AspenTech EU User Group 2003 - Oil Information Technology Journal

Aspen Tech European User Group 2003
clear opportunity to reduce gas lift to some wells to optimize production. This is a ‘classic’ coupled
problem between facility and well–related variables.
Pompano – division of labors
EXCEL HYSYS GAP
| | |
Excel model built by MDC to control
HYSYS and GAP opposite.
GAP is embedded into HYSYS and
shares variables.
Case history Valhall optimization
Topside
Petroleum Experts tool for well
performance optimization (GAP
aggregates Prosper models which
are ‘very popular’ in BP)
Valhall is a mature North Sea asset producing 100kbpd from 38 wells. Two HP/LP compression trains
with 7 stages of compression reflect incremental design creep and are perhaps the most complex facility
ever! Process from remote tie-back again demonstrates slugging. The study was performed by MDC.
Valhall benefited from a good infrastructure for online optimization with good bandwidth and metering.
MDC could observe operations in real time. The complex constrained compression system was the target
for optimization by making space for production in the compression system. A tax on produced CO2 was
also an issue. The optimization boiled down to 6 numbers that operators need to track. Success factors
included accurate asset models of wells and facilities based on tools of choice: Pipesim, HYSYS, Prosper
GAP. A consistent data set for validation was also key. Optimizing linked black oil and compositional
PVT models was a non trivial problem – models need to be ‘robust and available’ since optimization
cannot halt operations. People and process proved more complex issues due to the cross-disciplinary
nature and ownership of the project.
In fact both of these projects have ‘struggled’ to provide value for BP because of people issues. Valhall
optimization potential is significant–but operators have not taken much notice because the project did not
have buy-in from day one. The reaction tended to be ‘very interesting–now let me get on with my job 4 .’
The Valhall control room is moving onshore and it should be easier to manage optimization. Operators
need training and there needs to be a clear offering – perhaps a performance contract including a 2%
production hike. SimOpt needs to be delivered ‘at pace’, worldwide and for multiple assets. There is a
need for ‘plug & play’ models integrating technology and commercial issues. Steady-state and dynamic in
one model, Smarter optimization MINLP 5 , analytical derivatives and ‘models that answer questions that
have not been asked yet’ are aims.
Q&A
Q – Is the reservoir modeling community involved?
A – Yes, we talk to them although the timescale is different. It is an interesting dialog but we
really need a new tool to optimize across different disciplines and time scales.
4 Stenhouse also indicated that operators were concerned with potential gas blow-by as assets were driven closer to the edge.
5 Mixed Integer Nonlinear Programming.
Technology Watch Report 4 © 2003 The Data Room